def _get_testdata_file(name: str, download: bool = True) -> Optional[str]:
# Check pydicom local
data_path = Path(DATA_ROOT) / 'test_files'
matches = [m for m in data_path.rglob(name)]
if matches:
return os.fspath(matches[0])
# Check external data sources
fpath: Optional[str]
for lib, source in external_data_sources().items():
try:
fpath = source.get_path(name, dtype=DataTypes.DATASET)
except ValueError:
fpath = None
# For pydicom-data, check the hash against hashes.json
if lib == "pydicom-data":
if fpath and _check_data_hash(fpath):
return fpath
elif fpath:
return fpath
# Try online
if download:
for filename in get_url_map().keys():
if filename != name:
continue
try:
return os.fspath(data_path_with_download(filename))
except Exception:
warnings.warn(
f"A download failure occurred while attempting to "
f"retrieve {name}")
return None
def fetch_data_files():
"""Download missing test files to the local cache."""
cache = get_data_dir()
paths = {cache / fname: fname for fname in list(get_url_map().keys())}
error = []
for p in paths:
# Download missing files or files that don't match the hash
try:
data_path_with_download(p.name)
except Exception:
error.append(p.name)
if error:
raise RuntimeError(
"An error occurred downloading the following files: "
f"{', '.join(error)}")
def get_testdata_file(name: str) -> str:
"""Return an absolute path to the first matching dataset with filename
`name`.
.. versionadded:: 1.4
First searches the local *pydicom* data store, then any locally available
external sources, and finally the files available in the
pydicom/pydicom-data repository.
.. versionchanged:: 2.1
Modified to search locally available external data sources and the
pydicom/pydicom-data repository
Parameters
----------
name : str
The full file name (without path)
Returns
-------
str or None
The absolute path of the file if found, or ``None``.
"""
# Check pydicom local
data_path = Path(DATA_ROOT) / 'test_files'
matches = [m for m in data_path.rglob(name)]
if matches:
return os.fspath(matches[0])
# Check external data sources
for lib, source in external_data_sources().items():
try:
fpath = source.get_path(name, dtype=DataTypes.DATASET)
except ValueError:
fpath = None
# For pydicom-data, check the hash against hashes.json
if lib == "pydicom-data":
if fpath and _check_data_hash(fpath):
return fpath
elif fpath:
return fpath
# Try online
for filename in get_url_map().keys():
if filename == name:
try:
return os.fspath(data_path_with_download(filename))
except Exception:
warnings.warn(
f"A download failure occurred while attempting to "
f"retrieve {name}")
def get_testdata_file(name: str,
read: bool = False) -> Union[str, "Dataset", None]:
"""Return an absolute path to the first matching dataset with filename
`name`.
.. versionadded:: 1.4
First searches the local *pydicom* data store, then any locally available
external sources, and finally the files available in the
pydicom/pydicom-data repository.
.. versionchanged:: 2.1
Modified to search locally available external data sources and the
pydicom/pydicom-data repository
.. versionchanged:: 2.2
Added the `read` keyword parameter.
Parameters
----------
name : str
The full file name (without path)
read : bool, optional
If ``True`` then use :func:`~pydicom.filereader.dcmread` to read the
file and return the corresponding
:class:`~pydicom.dataset.FileDataset`. Default ``False``.
Returns
-------
str, pydicom.dataset.Dataset or None
The absolute path of the file if found, the dataset itself if `read` is
``True``, or ``None`` if the file is not found.
"""
from pydicom.filereader import dcmread
# Check pydicom local
data_path = Path(DATA_ROOT) / 'test_files'
matches = [m for m in data_path.rglob(name)]
if matches:
path = os.fspath(matches[0])
return dcmread(path, force=True) if read else path
# Check external data sources
for lib, source in external_data_sources().items():
try:
fpath = source.get_path(name, dtype=DataTypes.DATASET)
except ValueError:
fpath = None
# For pydicom-data, check the hash against hashes.json
if lib == "pydicom-data":
if fpath and _check_data_hash(fpath):
return dcmread(fpath, force=True) if read else fpath
elif fpath:
return dcmread(fpath, force=True) if read else fpath
# Try online
for filename in get_url_map().keys():
if filename == name:
try:
path = os.fspath(data_path_with_download(filename))
return dcmread(path, force=True) if read else path
except Exception:
warnings.warn(
f"A download failure occurred while attempting to "
f"retrieve {name}")
return None
def get_files(base: Union[str, os.PathLike],
pattern: str = "**/*",
dtype: int = DataTypes.DATASET) -> List[str]:
"""Return all matching file paths from the available data sources.
First searches the local *pydicom* data store, then any locally available
external sources, and finally the files available in the
pydicom/pydicom-data repository.
.. versionchanged: 2.1
Added the `dtype` keyword parameter, modified to search locally
available external data sources and the pydicom/pydicom-data repository
Parameters
----------
base : str or os.PathLike
Base directory to recursively search.
pattern : str, optional
The pattern to pass to :meth:`~pathlib.Path.glob`, default
(``'**/*'``).
dtype : int, optional
The type of data to search for when using an external source, one of:
* ``0`` - DICOM dataset
* ``1`` - Character set file
* ``2`` - Palette file
* ``3`` - DICOMDIR file
* ``4`` - JPEG file
Returns
-------
list of str
A list of absolute paths to matching files.
"""
base = Path(base)
# Search locally
files = [os.fspath(m) for m in base.glob(pattern)]
# Search external sources
for lib, source in external_data_sources().items():
fpaths = source.get_paths(pattern, dtype)
if lib == "pydicom-data":
# For pydicom-data, check the hash against hashes.json
fpaths = [p for p in fpaths if _check_data_hash(p)]
files.extend(fpaths)
# Search http://github.com/pydicom/pydicom-data or local cache
# To preserve backwards compatibility filter the downloaded files
# as if they are stored within DATA_ROOT/test_files/*.dcm
dummy_online_file_path_map = online_test_file_dummy_paths()
dummy_online_file_path_filtered = fnmatch.filter(
dummy_online_file_path_map.keys(), os.path.join(base, pattern))
download_names = [
os.fspath(dummy_online_file_path_map[dummy_path])
for dummy_path in dummy_online_file_path_filtered
]
real_online_file_paths = []
download_error = False
for filename in download_names:
try:
real_online_file_paths.append(
os.fspath(data_path_with_download(filename)))
except Exception:
download_error = True
files += real_online_file_paths
if download_error:
warnings.warn(
"One or more download failures occurred, the list of matching "
"file paths may be incomplete")
return files
